摘要
为有效预测和控制隧道开挖引起的地表超小变形,首先基于随机介质理论和MATLAB软件编写反分析程序,对地表沉降的实测数据进行拟合,反演断面收敛面积ΔA和地层影响角β,并研究ΔA与隧道埋深Z以及ΔA与tanβ的关系。其次,基于滚动预测的方法建立反向传播(back propagation,BP)神经网络,并预测隧道断面DM-1未来的地表超小变形值。结果表明:ΔA的取值较离散,随着隧道埋深Z的增加,ΔA呈逐渐减小趋势,地层损失也随之减小,当隧道埋深约为20.53 m时,收敛面积ΔA接近0。随着ΔA的逐渐增大,tanβ逐渐减小,当隧道埋深约为20.53 m时,tanβ约为0.447。通过滚动预测的方法建立的BP神经网络误差值较小,可以很好地预测隧道开挖引起的地表超小变形值,对类似的工程具有较好的指导意义。
In order to effectively predict and control the ultra-small surface deformation caused by tunnel excavation,firstly,based on the random medium theory and MATLAB software,a reverse analysis program was written to fit the measured data of surface subsidence.The convergence areaΔA and the formation influence angleβwere retrieved,and the relationship betweenΔA and the buried depth of the tunnel andΔA and tanβwas studied.Secondly,back propagation(BP)neural network was established based on rolling prediction method,and the future surface ultra-small deformation value of tunnel section DM-1 was predicted.The results show that the value ofΔA is relatively discrete.With the increase of the tunnel buried depth Z,ΔA shows a gradual decreasing trend,and the formation loss also decreases.When the tunnel buried depth is about 20.53 m,the convergence areaΔA is close to 0.With the gradual increase ofΔA,tanβgradually decreases.When the tunnel buried depth is about 20.53 m,tanβis about 0.447.The BP neural network established by rolling prediction method has a small error value,which can well predict the ultra-small surface deformation value caused by tunnel excavation,and has a good guiding significance for similar projects.
作者
杨三强
曹亚文
张丹
YANG San-qiang;CAO Ya-wen;ZHANG Dan(Hebei Civil Engineering Monitoring and Evaluation Technology Innovation Center,College of Civil Engineering,Hebei University,Baoding 071002,China)
出处
《科学技术与工程》
北大核心
2023年第1期369-375,共7页
Science Technology and Engineering
基金
国家自然科学基金(51808016)
河北省自然科学基金(E2018201106)。
关键词
浅埋隧道
地表超小变形
随机介质理论
BP神经网络
shallow buried tunnel
ultra-small surface deformation
random medium theory
BP neural network